A worker wearing an RGB camera on her head, recording actions through motion capture.
Every day, thousands of Indian workers strap cameras to their heads and record themselves cooking, cleaning, folding clothes, packing goods, and performing factory tasks. The footage helps train the next generation of AI-powered robots. For some, it is easy money. For others, it feels like they are helping build the machines that may one day replace them.
In a modest kitchen in Chennai, 25year old Nagireddy Sriramyachandra slices mangoes while a smartphone attached to her head records every movement.
She is not filming content for social media.
She is helping teach robots how humans perform everyday tasks.
For an hour of recording, she earns about 250 rupees, roughly $2.60. The footage is then uploaded to an AI data company, where it becomes part of a growing library used by technology firms developing humanoid robots and advanced artificial intelligence systems.
What appears to be an ordinary household chore is actually part of a rapidly expanding global industry.
Across India, thousands of workers are now being paid to generate what AI developers call “egocentric data” first person recordings that allow machines to observe the world through human eyes. Developers believe these videos will help robots learn how to navigate real environments, handle objects, and eventually perform many of the tasks humans do every day.
The work is taking place everywhere.
Some people record themselves inside their homes. Others wear cameras in factories, warehouses, workshops, and specially designed training studios. Workers are filmed folding towels, filling bottles, arranging objects, sorting products, cooking meals, and carrying out countless other routine activities. Motion sensors and head mounted cameras capture every detail of how the body moves.
For technology companies racing to build humanoid robots, this data has become incredibly valuable.
AI systems have already learned how to generate text, create images, and answer questions. Teaching machines to operate in the physical world, however, presents a much more difficult challenge. Robots must understand movement, coordination, timing, and interaction with objects.
That is where human workers come in.
The recordings provide real world examples of how people perform tasks that machines still struggle to master. Every hand movement, every adjustment, and every decision becomes part of a training dataset designed to make robots more capable.
Behind every robot learning to fold a towel or prepare food is a human being showing it exactly how the job is done.
India has emerged as a major hub for this kind of work.
The country already plays a central role in data labeling, content moderation, and AI training services. Now it is becoming a key supplier of physical world data as companies seek to build machines capable of performing both industrial and domestic tasks.
The timing is significant.
The humanoid robotics market is expected to expand dramatically in the coming decades. Some projections suggest more than one billion humanoid robots could be operating worldwide by 2050, primarily in industrial and commercial settings.
Yet the growth of the industry is also raising uncomfortable questions.
Many workers understand that the same technology they are helping train could eventually reduce the need for human labor.
One of them is Ponni, a flower garland maker in Bengaluru who has spent years working by the roadside. She participated in AI training projects but worries about what the future may hold for younger workers entering similar professions.
Experts say those concerns are not unfounded.
Automation has long promised efficiency and productivity, but it has also created anxiety about job displacement. In India, where hundreds of millions of people work in informal sectors, the debate is particularly sensitive. Government think tank NITI Aayog recently warned that discussions about AI often focus on white collar professions while paying less attention to the millions of manual and informal workers who could also be affected.
Public reaction online reflects that unease.
Many social media users have described the situation as workers “training their replacements,” while others argue that people have little choice but to take opportunities that provide immediate income. Some see the work as a practical way to earn money today. Others view it as a glimpse into a future where robots increasingly perform jobs once reserved for humans.
For now, the workers keep recording.
They film themselves cooking meals, folding clothes, organizing homes, and carrying out factory duties, often earning only a few dollars for their efforts.
The irony is difficult to ignore.
The people teaching machines how to move, think, and work are often the very people whose livelihoods could be transformed by the success of those machines.
And in kitchens, factories, and workshops across India, the future of robotics is being built one recorded task at a time.
